Grandin had posited in her book that many animals shared this characteristic --- that birds, for example, used this amazingly detailed visual memory to remember where they had hidden food caches, or that horses and cattle might panic at the sight of some intense visual signal, like a brightly colored T-shirt flapping in the wind, that most humans would not even register. While the authors of the article (Giorgio Vallortigara, Allan Snyder, Gisela Kaplan, Patrick Bateson, Nicola S. Clayton and Lesley J. Rogers) do not dispute that animal intelligence can involve this kind of specialized storage of massive amounts of sensory data, they disagree that all animal cognition is like that. They reference studies done on pigeon perception and intelligence, and found that pigeons (and other birds) are capable of both holistic, gestalt-type "global" perception and detailed, piecewise "local" perception. They also observe that animals with the capacity for prodigious spatial memory do not ever seem to experience sensory overload, nor do they show "deficits" in other areas of cognition or behavior*, both of which they identify with autism in humans. Other discrepancies they find between animal brains and autistic human brains are the intact mirror-neuron systems** that have been found in some animals, and the lack of a relationship between social behavior and intensive use of spatial cognition. These things lead them to decide that certain animals' highly-developed use of visual and spatial perception and memory are not so much features of a larger cognitive style as they are specific adaptations to each kind of animal's environment and lifestyle.
Temple Grandin has also written a response to this article (it's in the green box embedded in the article's text), in which she says that more study is needed to verify Vallortigara et al's claims about the cognition of spatially-gifted animals; specifically, she would want to test their abilities in areas related to but distinct from their particular skill, to see if that skill is "hyperspecific," or restricted to a particular context, which she claims would make it more like an autistic savant's skill.
*I don't think every autistic savant necessarily has "deficits" in skills that are not his or her savant skill; Daniel Tammet is one example I can cite right off as a savant who is also thought to have a high level of general intelligence. I think the defining characteristic of a savant would be the highly developed, hyperspecific "islets of ability," regardless of the person's level of ability in other areas. I think the stereotype of the "idiot savant" might lead people to notice savantism more in people with profound impairments in other areas, and also that those instruments we have to test intelligence are so narrowly specific to one type of intelligence that it is reasonable to suppose that it would tend more often to fail to identify intelligent people whose minds work in unusual ways, like savants.
**At some point I intend to do a post on mirror neurons. They're the Latest Thing in autism research, but I'm still not sure there's any there there.